JP6509928B2 - Variable resolution processing of frame type data - Google Patents

Description

  This application claims the benefit of US Provisional Patent Application No. 60 / 822,760, filed August 18, 2006, entitled "Variable Resolution Filtering," the entire contents of which are incorporated herein by reference. It is incorporated herein by reference as if recited.

  The present invention relates to signal processing, such as, for example, processing of audio signals.

  Many conventional signal processing techniques are of the frame type. In such techniques, a stream of data is divided into discrete frames, and the data in each such frame is usually processed very uniformly. In one example, the input speech signal is divided into frames of equal length. Each frame is then processed in a specific manner. A common processing parameter to be determined for each frame is the block length or, in other words, how many equal sized blocks the frame should be divided for processing purposes. The block length determines the resolution in the original domain (e.g. time for the audio signal) and in the frequency (or other transform) domain. More specifically, the shorter the block length, the higher the resolution in the original domain and the lower the resolution in the frequency domain.

  Speech signals often consist of quasi-stationary episodes, each of which contains many tone frequency components interrupted by significant transients. Thus, although individual frames of such speech signals will often include a small number of samples corresponding to transients, the majority of those samples correspond to the quasi-stationary portion of the signal.

  Since transients in the audio signal can be as short as a few samples, the block size used in a frame detected to contain transients is also ideally only a few samples, so that the filter The temporal resolution should be matched to the transient. Unfortunately, using different block sizes within the same frame is generally not practical. If every block in a frame with detected transients is only a few samples wide, the frequency resolution in the frame is extremely reduced and thus unsuitable for the remaining samples in the frame. That is, such other samples are quasi-stationary, so they should be processed with high frequency resolution, provided they are sufficiently far from transients. This contradiction has conventionally resulted in contradictory block sizes that are not optimal for transient samples or quasi-stationary samples in the same frame.

  FIG. 1 shows a block diagram of a conventional system for processing a frame of input samples 12. First, the sample group 12 is analyzed in the transient detector (transient detector) 14 to determine whether the frame includes a transient.

  Based on the detection, a window function is selected in module 16. In this regard, speech coding algorithms often use filterbanks with different time and frequency resolutions. One of the commonly used filter banks is an MDCT (Modified Discrete Cosine Transform) having an impulse response that can be expressed by the following basis function.

However, k = 0, 1,. . . , M−1, n = 0, 1,. . . , 2M−1, and w (n) are window functions of length 2M. For example, H. S. See Malvar, "Signal Processing with Lapped Transforms", Artech House, 1992 (referred to herein as Malvar).

  In this case, time-frequency resolution is determined by M, which may be referred to herein as block size. A large M means high frequency resolution but low time resolution, and a small M means high time resolution and low frequency resolution.

  For the purpose of implementing module 16 (as shown in FIG. 1), two block sizes are typically used in conventional coding algorithms. If no transients were detected in module 14, then a large block size is used which is implemented as a single block encompassing the entire frame. Alternatively, if a transient has been detected, a small block size implemented as a predetermined number of blocks containing the frame is used.

  The main window functions corresponding to these two block sizes are window function 30 (shown in FIG. 2 and denoted as WIN_LONG_LONG2LONG) and window function 40 (shown in FIG. 3 and denoted as WIN_SHORT_SHORT2SHORT) ). Three transition window functions, eg, windows, as a condition of full recovery (eg as described in Malvar), in order to be able to switch MDCT appropriately between these two main window functions Function 50 (shown in FIG. 16 and indicated as WIN_LONG_LONG2SHORT), window function 60 (shown in FIG. 5 and indicated as WIN_LONG_SHORT2LONG), and window function 70 (shown in FIG. 6 and indicated as WIN_LONG_SHORT2SHORT) It is required to use the It is noted that all three such transition window functions 50, 60 and 70 are used for long blocks (i.e. encompass the entire frame).

  Thus, in the prior art, a frame is assigned a single long block (and corresponding long windows 30, 50, 60 or 70) or a sequence of identical short blocks (and corresponding identical short windows 40) Be As each block is longer than the block-to-block spacing, the result is that each window has M new samples of the current block, such as the sequence 80 of the window function group shown in FIG. It is an overlapping sequence of long window and short window, including M samples in. For reference, in the drawing, the center of each block corresponding to the window function 30, 40, 50, 60 or 70 is designated as 31, 41, 51, 61 or 71, respectively.

  In such prior art, the window function for frames that do not include transients is based not only on the detection for such current frame performed by module 14 but also on similar detections performed for the previous and next frames. To be selected. That is, window functions 50, 60 and 70 are used as transition window functions between transient and non-transient frames.

  Referring again to FIG. 1, in module 17, the window function selected in module 16 is then applied to input samples 12 for the current frame (multiple times for transient frames). That is, for each block, the sample value is multiplied by the window function value corresponding to that block to obtain a set of weight values.

  These weights are then processed in module 19 using the selected window function to obtain output value 22. The specific type of processing performed in module 19 may vary depending on the desired application. For example, for speech signals, this processing may include analysis, coding, and / or enhancement.

  The present invention addresses this and other issues, inter alia, by using a plurality of different window functions within a frame that has been detected to contain a transient. In a preferred embodiment, the invention provides at least two levels of resolution within a single data frame with detected transients. More preferably, such multiple resolutions are obtained without changing the block size within the frame.

  As a result, for example, higher resolution may be used in the vicinity of transients and lower resolution in other parts of the frame. As used herein, the term "resolution", when used unconditionally, refers to resolution in the original (e.g., time) domain. Because the resolution in the frequency (or other transform) domain varies inversely with the resolution in the original domain, in these embodiments of the invention, the higher frequency (or other transforms) with respect to portions of the frame that do not include transients Region) resolution is obtained. Furthermore, by keeping the block size constant, the above advantages can generally be achieved without complicating the processing structure.

  Thus, in one respect, the invention is directed to the processing of frame type data to obtain a frame of data, an indicator indicating that a transient occurs in the frame, and the location of the transient in the frame. . Based on the indication of the transient, a block size is set for the frame, thereby effectively defining a plurality of blocks of the same size in the frame. And processing the frame of data by selecting different window functions for different ones of the plurality of identical sized blocks based on the location of the transient and applying the selected window function. .

  In a preferred embodiment, the blocks overlap each other, and each window function also overlaps each adjacent window function, preferably in such a way as to satisfy the conditions for full restoration. The above characteristics preferably apply to adjacent block and window functions in a frame, as well as to adjacent blocks and window functions in an adjacent frame.

  In any case, the window function is preferably selected to provide higher resolution within the identified one of the blocks of the same size, including the transient. Furthermore, this may be achieved by using, within the identified block, a transient window function that is narrower than the other of the window functions (e.g. samples within the block but outside the transient window function). Preferably, this is achieved by zeroing. In other words, while the width of the blocks is constant across the frame, the width of the window function is varied within those blocks, and the desired resolution for each block of the frame, as desired. Tradeoffs (eg, time / frequency) can be achieved.

  In this regard, the width of the window function can be defined in a number of different ways. For example, the window function may include the length of the non-zero portion of the window function, the length of the portion of the window function that exceeds a defined threshold, or the content (eg, energy) of the window function at a defined percentage. It can be defined as the length of the part of.

  Thus, the width of the window function can be varied by compressing or decompressing the standard form and then zeroing all samples within the block but not included in the compressed form. Alternatively, the widths can be varied, in part, by using various forms in which their energy is concentrated in smaller segments.

  In another respect, the present invention is directed to the processing of frame type data to obtain data of a frame, an indicator indicating that a transient occurs in the frame, and a position of the transient in the frame. The various window functions used in the frame are selected to process the frame of data by applying the selected window function so as to provide higher resolution in the area containing the transient.

  The above summary is intended only to briefly describe certain aspects of the present invention. A more complete understanding of the present invention can be obtained by reference to the claims and the following detailed description of the preferred embodiments, taken in conjunction with the accompanying drawings.

FIG. 1 is a block diagram of a conventional system for changing the temporal resolution in a data frame based on the presence of a transient. FIG. 2 shows a conventional window function for a frame having a long block size, if the frame is also sandwiched between two frames having a long block size. FIG. 3 shows a conventional window function for a frame having a short block size. FIG. 4 shows a conventional transition window function for a frame having a long block size, preceded by a frame having a long block size, and followed by a frame having a short block size. FIG. 5 illustrates a conventional transition window function for a frame having a long block size, preceded by a frame having a short block size, and followed by a frame having a long block size. FIG. 6 shows a conventional transition window function for a frame having a long block size preceded by a frame having a short block size and followed by a frame having a short block size. FIG. 7 shows a sequence of an exemplary conventional window function group in which a frame including a transient touches two frames not including a transient on both sides. FIG. 8 shows a brief window function WIN_SHORT_BRIEF2BRIEF according to a representative embodiment of the present invention. FIG. 9 shows a transition window function WIN_SHORT_SHORT2BRIEF according to a representative embodiment of the present invention. FIG. 10 shows a transition window function WIN_SHORT_BRIEF2SHORT according to a representative embodiment of the present invention. FIG. 11 shows a transition window function WIN_LONG_LONG2BRIEF according to a representative embodiment of the present invention. FIG. 12 shows a transition window function WIN_LONG_BRIEF2LONG according to a representative embodiment of the present invention. FIG. 13 shows a transition window function WIN_LONG_BRIEF2BRIEF according to a representative embodiment of the present invention. FIG. 14 shows a transition window function WIN_LONG_SHORT2BRIEF according to a representative embodiment of the present invention. FIG. 15 shows the transition window function WIN_LONG_BRIEF2 SHORT according to a representative embodiment of the present invention. FIG. 16 is a flow chart illustrating a technique for selecting a window function according to a representative embodiment of the present invention. FIG. 17 shows a first exemplary window function sequence according to the invention. FIG. 18 shows a second exemplary window function sequence according to the invention. FIG. 19 shows a third exemplary window function sequence according to the present invention. FIG. 20 shows a fourth exemplary window function sequence according to the invention.

  The main object of the present invention is to improve the window function selection element 16 of the conventional system. One feature of the present invention is the introduction of a new "brief window function", for example the window function 100 as shown in FIG. This window function 100, denoted as WIN_SHORT_BRIEF2BRIEF, is intended to be used within a short block, similar to the conventional window function WIN_SHORT_SHORT2SHORT 40. In fact, as described in more detail below, in the preferred embodiment of the present invention, the window function WIN_SHORT_BRIEF2BRIEF 100 is intended for use in the same frame as the window function WIN_SHORT_SHORT2SHORT 40.

  However, unlike the conventional window function, the brief window function 100 occupies it by using multiple leading and lagging zero weights to improve its temporal resolution (end points 102 and 103 in FIG. 8) Use only the central part of the total length of the block) for signal shaping. For example, if short block length = 256 samples, the brief window function 100 is preferably non-zero within the center 160 samples (in this case it is also symmetrical), the first of such samples 16 and the last 16 overlap with each transition window function adjacent to the brief window function 100, with the first 48 samples and the last 48 samples of the window zero weighted.

  In the preferred embodiment of the present invention, this brief window function 100 is used only when transient samples in the speech frame are detected (e.g. in the block containing the transient samples) while the normal short window function ( For example, a conventional window function 40), or one of the new transfer functions provided by the present invention, is applied to the quasi-stationary samples in the rest of the frame. This offers the following possibilities compared to the prior art:

  • A significantly larger block size can be used, in which case the frequency resolution for quasi-stationary samples is improved while the time resolution around transient samples remains unchanged.

  The block size can be left unchanged, in which case the frequency resolution for quasi-stationary samples is unchanged, while the time resolution around transient samples is improved.

• A slightly larger block size can be used, in which case the frequency resolution for quasi-stationary samples is slightly improved, while the time resolution around transient samples is also slightly improved.
With any of the above options, eclectic solutions that address conflicting time-frequency resolution requirements are improved.

An additional transition window function has been introduced to make it easier to use the window function 100 with the other two main window functions, ie, WIN_LONG_LONG2LONG 30 and WIN_SHORT_SHORT2SHORT 40 (eg, to satisfy the conditions of a full restore) Is preferred. An example of such a transition window function is given below. First, however, it should be mentioned that the following technical terms are generally used in the present disclosure.
WIN_BlockLength_PriorWF2SubsWF, where BlockLength indicates the length of the block occupied by the current window function (eg, long or short), and PriorWF indicates the type of window function in the immediately preceding block (eg, long, short or brief) Identify, SubsWF identifies the type of window function (eg, long, short or brief) in the block immediately following.

  The transition window function 110 (shown in FIG. 9 and labeled as WIN_SHORT_SHORT2BRIEF) is used within the short block immediately before another short block that contains a transient. That is, it is immediately before the window function WIN_SHORT_BRIEF2BRIEF100.

  The window function 120 (shown in FIG. 10 and labeled as WIN_SHORT_BRIEF2 SHORT) is used in the short block immediately after another short block that includes a transient. That is, it is immediately after the window function WIN_SHORT_BRIEF2BRIEF100.

  The window function 130 (shown in FIG. 11 and denoted as WIN_LONG_LONG2BRIEF) is used in the long block immediately before the short block containing the transient. That is, it includes the frame immediately prior to the next frame that contains the transient, which is in the very first block of the next frame.

  The window function 140 (shown in FIG. 12 and denoted as WIN_LONG_BRIEF2LONG) is used in the long block immediately after the short block containing the transient. That is, it includes the frame immediately following the preceding frame that contains the transient, which is in the very last block of the preceding frame.

  The window function 150 (shown in FIG. 13 and denoted as WIN_LONG_BRIEF2BRIEF) is used in a long block sandwiched between two short blocks containing a transient. That is, it includes the frame between two frames that includes a transient, the transient in the previous frame is in its very last block, and the transient in the next frame is in its very first block.

  The window function 160 (shown in FIG. 14 and denoted as WIN_LONG_SHORT2BRIEF) is used within a long block that contains a frame between two frames that contain a transient, and the transient in the previous frame is not that very last block Somewhere in the next frame, the transient is in its very first block.

  The window function 170 (shown in FIG. 15 and denoted as WIN_LONG_BRIEF2SHORT) is used in a long block that contains a frame between two frames that contain a transient, and the transient in the previous frame is that very last block Yes, the transient in the next frame is somewhere other than its very first block.

  In each case, the window function is preferably designed to overlap the adjacent window functions on both sides in such a way as to satisfy the condition of perfect reconstruction. Specific examples of window functions that can be used are given below.

  FIG. 16 is a flow chart illustrating a technique for selecting a window function according to a representative embodiment of the present invention. In general, the entire process shown in FIG. 16 will be automated (eg, implemented in software, firmware, dedicated hardware, or any combination thereof).

  First, at step 202, a frame of data is obtained. In this regard, a wide variety of different types of data may be processed in accordance with the present invention. Throughout this disclosure, data is often assumed to correspond to speech signals. However, this should not be interpreted as limiting and the data obtained may rather be image signals, video signals or heat, pressure, radiation, movement, distance, any biological function, weather and It may represent any other physical phenomenon, such as a signal representing any geological phenomenon.

  Also, it should be noted that the data frame may be defined by the source of data (eg, as in the case where data is being received through the communication channel). Alternatively, for example, data may be received in a continuous stream and segmented into frames (eg, internally) for processing purposes. In any case, the process is (although not exclusively) particularly applicable to data separated into individual frames. As indicated above, frame-type processing allows individual parts of the entire data stream to be processed uniformly at some specific points.

  In a preferred embodiment of the invention, each frame has a uniform block size. In this regard, blocks are preferably defined as basic signal processing units for the frame. For example, define data in a frame (eg, using discrete cosine transform or fast Fourier transform), from the original domain (eg, time domain for speech signals) to the frequency domain, or by a set of orthogonal functions If it is to be transformed (for example, in the signal processing module 19) into any other transform domain to be performed, it is preferable that the transformation in the transform domain and all subsequent processing be performed separately for each block. .

  Thus, in a preferred embodiment, the frame may be contained in a single block, or alternatively may be contained in multiple blocks of the same size. As in the prior art, the block size is more preferably only two: a large block size encompassing the entire frame and a small block size resulting in a plurality of contiguous blocks uniformly distributed throughout the frame .

  Furthermore, to address the boundary problems that would otherwise occur, as in the prior art, the blocks of the present invention preferably overlap each other (e.g., in such a way as to satisfy the conditions of perfect reconstruction). Conceptually, each block can be considered to contain a number of core samples to be processed next (e.g., in module 19) and a number of boundary samples adjacent to such core samples. In the preferred embodiment, the core sample is a new sample in the sequence and the boundary sample is a historical sample from the previous block. On the other hand, the frames are preferably continuous and non-overlapping. As a result, the block at the beginning of the frame overlaps the previous frame. For a frame contained in a single block, the single block overlaps the entire preceding frame.

  In addition to obtaining the data frame itself, step 202 also obtains a transient indicator (e.g., from transient detector 14). In a preferred embodiment, the transient label obtained indicates whether a transient is present in the current frame, and if so where in the frame. Preferably, if more than one transient is detected in the current frame, the location of each such transient is obtained (e.g. identified by transient detector 14 and then received from transient detector 14) . To simplify the description, in the present disclosure, it may be assumed that only one transient (if any) is detected in each frame without loss of generality.

  The detection of the actual transient can be performed, for example, using any existing technology. Transients usually manifest themselves as spikes in high frequency components for a very short time, so that this can be detected. In any case, in many cases a threshold is defined, below which signal activity will be considered as transient.

  Referring again to FIG. 16, in step 203 it is determined whether there is a transient in the current frame. This determination is preferably straightforward based on the information provided by the transient detector 14. If there is no transient, processing proceeds to step 205. If there is a transient, processing proceeds to step 210.

  At step 205, the block size is set based on the determination that the current frame does not contain a transient. In a preferred embodiment, in such a case, a single block is used to encompass the entire frame. More preferably, this block includes all samples as core samples in the current frame, and some or all samples from the preceding frame (s). Exemplary block sizes are 2,048 samples, ie 1,024 core samples from the previous frame (frame size is also 1,024 samples) and 1,024 samples.

  Next, in step 207, a window function is selected for the current frame (assuming that a single block covers the entire frame). In a preferred embodiment, this step includes the evaluation of the immediately preceding and succeeding frames / blocks. Due to the increasing number of window functions, the determination of the appropriate window sequence is typically somewhat complicated compared to the prior art, but the basic principles are relatively straightforward and clear. Specifically, a long window function is selected, the specific form of which depends on the presence and location of the transient in the previous and next frames. The specific selection is preferably performed as follows.

Once the appropriate window function is selected at step 207, it is applied at step 17 (e.g., as described above).

  On the other hand, if it was determined in step 203 that a transient was present in the current frame, the process would proceed to step 210 to set the block size to "small". An example for a frame size of 1,024 samples is a block size of 256 samples, i.e. 128 core samples and 128 samples overlapping with the preceding block (so that the frame is contained in 8 blocks). Although this embodiment contemplates a single block size for each of two possible situations (transient / no transient), in other embodiments different block sizes may be based on any desired criteria. It should be noted that one may choose, and the frame may consist of blocks of various sizes.

  In any case, once the block size is established, the process proceeds to step 212 where various window functions are selected for various blocks in the current frame. Since it is known that the current frame contains at least one transient, the WIN_SHORT_BRIEF2BRIEF window function 100 will be used at least once (at the identified position (s)). More preferably, a sequence of brief window functions and short window functions is selected for the short block of the current frame according to the following principle.

  To improve the temporal resolution of the block, WIN_SHORT_BRIEF2BRIEF is applied to the samples in the block where the transient occurs.

  The window function for the window immediately before the window containing the transient has a specification of the form “... 2BRIEF”.

  The window function for the window immediately after the window including the transient has a specification of the form “... _BRIEF 2.

  Therefore, any combination of the following window functions is acceptable.

The specific procedure written in C for placing a window function in a frame is described in the section entitled "Window Function Selection Routine" of this disclosure. However, in general, the pre-transient transition window function is selected from the left column of the above table based on its own block length (long or short) and the nature of the window function (long, short or brief) immediately preceding it. Be done. Similarly, a post-transient transition window function is selected from the right column of the above table based on its own block length (long or short) and the nature of the window function (long, short or brief) immediately following it. (I) before transient (the block immediately before transient block), (ii) transient (the block including transient), (iii) after transient (the block immediately after transient block), (iv) the first block of the frame, or v) The last block of the frame, preferably not each window in the current frame, is assigned a window function of WIN_SHORT_SHORT2SHORT 40. The first block in the current frame is preferably assigned a window function of WIN_SHORT_PriorWF2 SHORT (assuming that it is not a pre-transient, post-transient or post-transient block), the last block of the current frame is (transient Preferably, a window function of WIN_SHORT_SHORT2SubsWF is assigned (assuming that it is not pre-transient or post-transient block).

  Once step 212 is complete, the process proceeds to step 17 to apply the selected window function. When step 17 is completed, the process returns to step 202 to process the next frame.

  It should be understood that the flowchart shown in FIG. 16 is inherently conceptual in nature. In fact, as indicated above, the choice of window function (s) for one frame will typically be influenced by the presence and location of transients in adjacent frames. Thus, the selection of window function (s) for one frame may be performed simultaneously with the selection of window function (s) for one or more other frame groups. Preferably, at least the selection for the current frame is made in anticipation of the window function that will be applied to the next frame.

  It is noted that portion 216 of the flowchart shown in FIG. 16 can be used in place of the corresponding processing performed in the conventional window function selection module 16 (in FIG. 1). As a result, the techniques of the present invention can typically be easily incorporated as an improvement over prior art systems.

  Speech coding / decoding is one application where the present invention may be used. Within such a system, the encoder typically allows the decoder to decode the frame using the same window function as the window function that it used to encode the current frame Inform the decoder. With regard to the prior art, in order to achieve this purpose, generally only one window function index need be sent to the decoder for the following reasons.

  For frames without detected transients, only one of the long window functions is used, so it needs to be known to the decoder.

  For a frame that contains more than one transient, all window functions in the frame are the same, ie, WIN_SHORT_SHORT2SHORT 40. Such a frame (embedded between two frames without detected transients) is shown in FIG.

  The above comments also apply to the technology of the present invention. That is, only one window function index need be sent to the decoder in order for the decoder to decode the frame using the same window functions as the encoder. This is due to the following reasons.

  -For frames that do not have a detected transient, this technique only needs to inform the decoder which one of the long window function group is used.

-For frames with detected transients, the encoder will decode if there is a transient in the first block of the current frame and if there is a transient in the first block of the next frame The sequence of window functions for the entire frame can then be determined using the procedures described herein. In the preferred embodiment, the WIN_SHORT_BRIEF2BRIEF window function 100 is used for blocks with transients, so the following terminology may be used to convey this information.
WIN_SHORT_CurrentSubs, where Current (SHORT = No, BRIEF = Yes) identifies whether a transient is present in the first block of the current frame, Subs (SHORT = No, BRIEF = Yes) is for the next frame Identify if there is a transient in the first block. For example, WIN_SHORT_BRIEF2BRIEF indicates that a transient exists in the first block of the current frame and the first block of the next frame, and WIN_SHORT_BRIEF2SHORT indicates that the transient exists in the first block of the current frame, but Indicates that it does not exist in the first block.

  Figures 17 to 20 show an exemplary window function sequence according to the invention. In each such sequence, a frame having at least one transient (as indicated by the WIN_SHORT_BRIEF2BRIEF window function 100) is enclosed in two frames that do not include transients on both sides. However, these examples should not be construed as limiting, and continuous frames, each having at least one transient, are also acceptable.

Exemplary Implementation One of the widely used window functions is the following sine function.

If the long window function is M = L, the short window function is M = S, and the brief window function is M = B, where L> S> B, the following window functions can be defined.

  A preferred set of window length parameters is L = 1024, S = 128, and B = 32. However, other parameters may be used instead.

System Environment In general, all systems, methods, and techniques described herein, except where expressly indicated to the contrary, include one or more programmable general purpose computing devices It is feasible by use of. Such devices will typically include, for example, at least a portion of the following components interconnected by a common bus or the like: one or more central processing units (CPU); read only memory ( Random access memory (RAM); (for example, using a hardwired connection such as serial port, parallel port, USB connection, or firewire connection, or using a wireless protocol such as Bluetooth or 802.11 protocol ) Input / output software and circuitry to interface with other devices (eg, hardwired connections such as Ethernet cards, or Code Division Multiple Access (CDMA), Global System Four) Mobile Communications (GSM (registered trademark) (Using a wireless protocol such as Bluetooth, 802.11 protocol, or any other cellular or non-cellular system), and in many embodiments of the present invention, then connect to the Internet or any other network Software and circuitry for connecting one or more networks; displays (eg, cathode ray tube displays, liquid crystal displays, organic light emitting displays, polymer light emitting displays, or any other thin film display); Other output devices (eg, one or more speakers, a headphone set, and a printer); one or more input devices (eg, a mouse, a touch pad, a tablet, a touch sensor display or other pointing device, a keyboard, a keypad) , Microphones, and scanners); mass storage (Eg, hard disk drive); real time clock; removable storage read / write device (eg, for reading from and writing to RAM, magnetic disk, magnetic tape, magneto-optical disk, optical disk etc); Modem) for sending faxes or connecting to the Internet or any other computer network via a dial-up connection. In operation, steps to carry out the above methods and functions to the extent performed by such a general purpose computer are typically first stored in a mass storage device (eg hard disk) and downloaded to RAM , Is executed by the CPU of the RAM. However, in some cases, the process is first stored in RAM or ROM.

  Suitable devices for use in the practice of the present invention can be obtained from various manufacturers. In various embodiments, different types of devices are used, depending on the size and complexity of the task. Suitable devices include smaller computers, such as a main computer, a multiprocessor computer, a workstation, a personal computer, and a PDA, a wireless telephone, or any other equipment or device, which are stand alone Also, it may be hardwired to the network or wirelessly connected to the network.

  Furthermore, while general purpose programmable devices have been described above, in other embodiments, one or more dedicated processors or computers may be used instead (or in addition). In general, unless stated otherwise, all the functions mentioned above can be implemented in software, hardware, firmware or any combination thereof, this particular implementation being a known technology It should be noted that the choice is made based on tradeoffs. More specifically, as the person skilled in the art will easily understand, if the functions mentioned above are implemented in a fixed, preset or logical manner, then they are programming (e.g. Software or firmware), appropriate arrangement of logic elements (hardware), or any combination of the two may be achieved.

  It should be understood that the present invention also relates to a machine readable medium on which program instructions for performing the methods and functions of the present invention are stored. Such media include, by way of example, magnetic disks, magnetic tapes, optically readable media such as CD ROMs and DVD ROMs, or semiconductor memories such as PCMCIA cards, various memory cards, USB memory devices, etc. Is included. In any case, the medium may be in the form of a portable item such as a small disk drive or small disk, diskette, cassette, cartridge, card, stick, etc. or be provided in a hard disk drive, computer or other device. It may be in the form of a relatively larger or fixed item such as a ROM or RAM.

  In the above description, the emphasis is mainly on computer computers and devices. However, it should be understood that any other computing device or other type of device may be used, such as a device using any combination of electronic, optical, biological and chemical processing. is there.

Additional Considerations Above, several different embodiments of the present invention have been described, each including specific features. However, as one of ordinary skill in the art would appreciate, the features described with respect to the description of any one embodiment are not limited to that embodiment, but are included in various combinations in any of the other embodiments, and It is intended that it may be / or arranged.

  Similarly, in the above description, functions may be attributed to particular modules or elements. However, functions may generally be redistributed, as desired, among the various modules or elements, and in some cases, the need for particular elements or modules may be completely eliminated and / or of new elements or modules. It will need to be added. The exact distribution of functions is preferably performed according to known technology tradeoffs with reference to the specific embodiments of the present invention, as will be appreciated by those skilled in the art.

  Thus, while the present invention has been described in detail in connection with exemplary embodiments thereof and the accompanying drawings, it is understood that various adaptations and modifications of the invention can be achieved without departing from the spirit and scope of the invention. It should be clear to the person skilled in the art. Accordingly, the invention is not limited to the exact embodiments shown in the drawings and described above. Rather, it is intended that all such variations which do not depart from the spirit of the present invention be considered as limited within the scope thereof and only by the claims appended hereto.

Window function selection routine

However:

Claims (12)

A computer, a computer program for executing a method of processing a frame-type data, the method comprising:
Acquiring (a) data of the frame,
(B) setting a block size for the frame, thereby defining a plurality of blocks of the same size in the frame;
(C) selecting different window functions for different ones of the plurality of blocks of the same size;
(D) processing the frame of data by applying the window function selected in step (c);
In step (c), a brief window function is selected for the block containing the transient, the brief window function is narrow compared to the other window functions in the frame containing the transient, and the energy in the smaller part of the block More of are concentrated,
The frame, the data stream is divided, one of the sequence of frames do not overlap successive same length discrete of
A computer program characterized in that the discrete continuous non-overlapping frames are processed uniformly.   The computer program according to claim 1, characterized in that the window functions overlap one another in such a way that the conditions for a complete restoration are satisfied.   A method according to claim 1 or 2, characterized in that the method comprises repeating steps (a) to (d) multiple times for different frames of the frame of discrete continuous non-overlapping data. The computer program described in.   The window function selected in step (c) further includes at least one of a standard transient frame window function, (i) a pre-transient transition window function, and (ii) a post-transient transition window function. The computer program according to any one of claims 1 to 3.   A computer program according to any of the preceding claims, wherein the brief window function starts with a first sequence of leading zero weights and ends with a second sequence of lagging zero weights.   4. The method according to claim 1, further comprising the step of transforming the data in the frame from the original region to a second region by separately transforming the data in each block of the frame. The computer program according to any one of 5.   The brief window function is defined by a first segment of zero weight followed by a weight defined by a first sine function followed by a segment of equal weight followed by a second sine function 7. A computer program according to any one of the preceding claims, characterized in that it comprises a second weight followed by a second weight of zero weight. The weight of the brief window function, for S> B,

The computer program according to any one of claims 1 to 7, characterized in that   9. The computer program according to claim 8, wherein S = 128 and B = 32.   A method according to claim 1, characterized in that the method comprises the step of assigning a single window function covering a second frame from the sequence of discrete continuous non-overlapping frames of the same length. The computer program according to any one of 9. The frame is covered by eight blocks, the second frame is covered by a single block,
The computer program of claim 10, wherein a window function is assigned to each of the blocks.   For each of the given frames in the sequence of discrete continuous non-overlapping frames of the same length, the block size is set based on whether or not there is a transient in the given frame. The computer program according to any one of claims 1 to 11, wherein

Images